Semantic Annotation for Microblog Topics Using Wikipedia Temporal Information

Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 97–106, Lisbon, Portugal, 17-21 September 2015. c⃝2015 Association for Computational Linguistics. Semantic Annotation for Microblog Topics Using Wikipedia Temporal Information Tuan Tran L3S Research Center Hannover, Germany ttran@L3S.de Nam Khanh Tran L3S Research Center Hannover, Germany ntran@L3S.de Asmelash Teka Hadgu L3S Research Center Hannover, Germany teka@L3S.de Robert J¨aschke L3S Research Center Hannover, Germany jaeschke@L3S.de Abstract Trending topics in microblogs such as Twitter are valuable resources to under- stand social aspects of real-world events. To enable deep analyses of such trends, se- mantic annotation is an effective approach; yet the problem of annotating microblog trending topics is largely unexplored by the research community. In this work, we tackle the problem of mapping trending Twitter topics to entities from Wikipedia. We propose a novel model that comple- ments traditional text-based approaches by rewarding entities that exhibit a high tem- poral correlation with topics during their burst time period. By exploiting temporal information from the Wikipedia edit his- tory and page view logs, we have improved the annotation performance by 17-28%, as compared to the competitive baselines. 1 Introduction With the proliferation of microblogging and its wide influence on how information is shared and digested, the studying of microblog sites has gained interest in recent NLP research. Several ap- proaches have been proposed to enable a deep un- derstanding of information on Twitter. An emerg- ing approach is to use semantic annotation tech- niques, for instance by mapping Twitter informa- tion snippets to canonical entities in a knowledge base or to Wikipedia (Meij et al., 2012; Guo et al., 2013), or by revisiting NLP tasks in the Twitter do- main (Owoputi et al., 2013; Ritter et al., 2011). Much of the existing work focuses on annotating a single Twitter message (tweet). However, infor- mation in Twitter is rarely digested in isolation, but rather in a collective manner, with the adoption of special mechanisms such as hashtags. When put together, the unprecedentedly massive adoption of Hard to believe anyone can do worse than Russia in #Sochi. Brazil seems to be trying pre;y hard though! spor=ngnews.com… #sochi Sochi 2014: Record number of posi=ve tests -­‐ SkySports: q.gs/6nbAA #Sochi Sea Port. What a beau=ful site! #Russia 2014_Winter_Olympics Port_of_Sochi Figure 1: Example of trending hashtag annota- tion. During the 2014 Winter Olympics, the hash- tag ‘#sochi’ had a different meaning. a hashtag within a short time period can lead to bursts and often reflect trending social attention. Understanding the meaning of trending hashtags offers a valuable opportunity for various applica- tions and studies, such as viral marketing, social behavior analysis, recommendation, etc. Unfor- tunately, the task of hashtag annotation has been largely unexplored so far. In this paper, we study the problem of annotat- ing trending hashtags on Twitter by entities de- rived from Wikipedia. Instead of establishing a static semantic connection between hashtags and entities, we are interested in dynamically linking the hashtags to entities that are closest to the un- derlying topics …

Content truncated to 500 words